Repository navigation
Conversation
| self.task_id = int(task_id) | ||
| self.task_type_id = int(task_type_id) | ||
| self.task_type = task_type | ||
| self.dataset_id = int(data_set_id) |
There was a problem hiding this comment.
can we call this "source_data", in co-ordinance with the task descriptions (hidden in the XML)
| self.estimation_procedure["data_splits_url"] = data_splits_url | ||
| self.estimation_procedure["parameters"] = estimation_parameters | ||
| # | ||
| self.estimation_parameters = estimation_parameters |
There was a problem hiding this comment.
I would rather store the estimation procedure id (not available in api right now, let's add it)
There was a problem hiding this comment.
@janvanrijn I'm afraid that won't be backwards compatible... But we can probably transit at some point to store something different in the cache. Do you want to open a new issue on the OpenML issue tracker?
| """Download dataset associated with task""" | ||
| return datasets.get_dataset(self.dataset_id) | ||
|
|
||
| def push_tag(self, tag): |
There was a problem hiding this comment.
didn't we have general tag functions?
There was a problem hiding this comment.
Otherwise we should create them. (and not duplicate tag functions code for each entity)
There was a problem hiding this comment.
Great remark. I think we should do this in a new PR, though, as it is not directly related to this PR. I opened issue #549 for this.
Codecov Report
@@ Coverage Diff @@
## develop #544 +/- ##
===========================================
- Coverage 90.07% 90.04% -0.03%
===========================================
Files 32 32
Lines 2811 2844 +33
===========================================
+ Hits 2532 2561 +29
- Misses 279 283 +4
Continue to review full report at Codecov.
|
5 participants
What does this PR implement/fix? Explain your changes.
It generalizes tasks so that you can handle different tasks differently.
Right now every task was assumed to be a classification task
How should this PR be tested?
Standard unit test